|
The technology of License Plate Recognition (LPR) is an increasingly mature. At present, majority of countries and regions have their own license plate recognition system, and all of them have high recognition rate, but the growing complexity transport system and the uncertainty environment bring higher requirement to the technology of LPR, so it has certain practical significance to research on the technology. In this paper, the key technique of LPR are researched and discussed on the basis of the former works, and also have been developed.In the gray processing for the color license plate image, according to the color of the character and background in the license plate, this paper researches the correlation between the color image and its three color components, and proposes to take the values of color components as the values of gray image, which can solve the shortage brought by using a single gray processing. Based on both mathematical morphology and the character characteristic, this paper uses the combination algorithm to locate the license plate.In character segmentation, this paper use image fuzzy-enhancement to deal with the license plate image before image binary firstly, then, use the mathematical morphology to remove noise effectively, at the last, use projection and the characters characteristics to segment the character.In character recognition, in order to recognize the character with a certain rotation angle and different size, this paper use Circle projection algorithm. Compared to other recognize algorithms, it can not only recognize the character with a certain rotation angle, but also needn’t to do character normalize, so it can speed up the character reorganization. Circle projection has two steps:the Central location and the image feature extraction. In the first step, although these are many methods to locate the circle central, most of them have some problems, such as the central location accuracy isn’t enough or the speed of them is slow, this paper proposes a new central location-based on the theory of the triangle and the quadrilateral, it can locate the center fast and accurately, and solve the difficult point of central location, which expand the use range of circle projection. In the feature extraction of image, this paper introduces a new method to get the feature of the characters, its projection value can save more information of image without increasing any complexity, which can enhance the character recognition rate.
|